Data Import Splitting Feature test report (Orchid) ocp3 + retesting Poppy FSF and RTR
Overview
The Data Import Task Force (DITF) implements a feature that splits large input MARC files into smaller ones, resulting in smaller jobs, so that the big files could be imported and be imported consistently. This document contains 1. Test with 1, 2, and 3 tenants' concurrent jobs with configurations the results of performance tests on the feature and also an analysis the feature's performance with respect to the baseline tests. The following Jiras were implemented.
https://folio-org.atlassian.net/browse/PERF-644 https://folio-org.atlassian.net/browse/PERF-645 https://folio-org.atlassian.net/browse/PERF-647 https://folio-org.atlassian.net/browse/PERF-646 https://folio-org.atlassian.net/browse/PERF-671
Summary
The file-splitting feature is stable and offers more robustness to Data Import jobs even with the current infrastructure configuration. If there were failures, it's easier now to find the exact failed records to take actions on them.
No stuck jobs in all tests performed.
There were errors (see below) in some partial jobs, but they still completed so the entire job status is "Completed with error".
Both of kinds of imports, create and update MARC BIBs worked well with this file-splitting feature enabled and also disabled.
There is no performance degradations, jobs not getting slower, on single-tenant imports. On multi-tenants imports, performance is be a little better
Duration for DI correlates with number of the records imported (100k records- 38 min, 250k - 1 hour 32 min, 500k - 3 hours 29 min).
Multitenant DI could be performed successfully for up to 9 jobs in parallel. If jobs are big they will start one by one in order for each tenant but processed in parallel on 3 tenants. Small DI (1 record) could be finished faster not in order.
No memory leak is suspected for all of the modules.
Average CPU usage for mod-inventory -was 144%, mod-di-converter-storage was about 107%, and for all other modules did not exceed 100 %. We can observe spikes in CPU usage of mod-data-import at the beginning of the Data Import jobs up to 260%. Big improvement over the previous version (without file-splitting) for 500K imports where mod-di-converter-storage's CPU utilization was 462% and other modules were above 100% and up to 150%.
Approximately DB CPU usage is up to 95%.
Recommendations and Jiras
One record on one tenant could be discarded with error: io.netty.channel.StacklessClosedChannelException. https://folio-org.atlassian.net/browse/MODDATAIMP-748 Reproduces in both cases with and without splitting feature enabled in at least 30% of test runs with 500k record files and multitenant testing.
During the new Data Import splitting feature testing, items for update were discarded with the error: io.vertx.core.impl.NoStackTraceThrowable: Cannot get actual Item by id: org.folio.inventory.exceptions.InternalServerErrorException: Access for user 'data-import-system-user' (f3486d35-f7f7-4a69-bcd0-d8e5a35cb292) requires permission: inventory-storage.items.item.get. Less than 1% of records could be discarded due to missing permission for 'data-import-system-user'. Permission was not added automatically during the service deployment. I added permission manually to the database and the error does not occur anymore.https://folio-org.atlassian.net/browse/MODDATAIMP-930
UI issue, when canceled or completed with error Job progress bar cannot be deleted from the screen. https://folio-org.atlassian.net/browse/MODDATAIMP-929
Usage:
Should not use less than 1000 for RECORDS_PER_SPLIT_FILE. The system is stable enough to ingest 1000 records consistently and smaller amounts will incur more overheads, resulting in longer jobs' durations. CPU utilization for mod-di-converter-storage for 500 RECORDS_PER_SPLIT_FILE(RPSF) = 160%, for 1000RPSF =180%, for 5K RPSF =380% and for 10K RPSF =433%, so in the case of selecting configurations 5K or 10K we recommend to add more CPU to mod-di-converter-storage service.
When toggling the file-splitting feature, mod-source-record-storage, mod-source-record-manager's tasks need to be restarted.
Keep in mind about the Kafka broker's disk size (as bigger jobs - up to 500K - can be run now), consecutive jobs may use up the disk quickly because the messages' retention time currently is set at 8 hours. For example with 300GB disk size, consecutive jobs of 250K, 500K, 500K sizes will exhaust the disk.
More CPU could be allocated to mod-inventory and mod-di-converter-storage
Results
Test # |
| Splitting Feature Enabled | Results | Splitting Feature Disabled | Results | Before Splitting Feature Deployed | Results | |
|---|---|---|---|---|---|---|---|---|
1 | 100K MARC BIB Create | PTF - Create 2 | 37 min -39 min | Completed | 40 min | Completed | 32-33 minutes | Completed |
1 | 250K MARC BIB Create | PTF - Create 2 | 1 hour 32 min | Completed | 1 hour 41 min | Completed | 1 hour 33 min - 1 hour 57 min | Completed |
1 | 500K MARC BIB Create | PTF - Create 2 | 3 hours 29 min | Completed* | 3 hours 55 min | Completed | 3 hours 33 min | Completed |
2 | Multitenant MARC Create (100k, 50k, and 1 record) | PTF - Create 2 | 2 hours 40 min | Completed* | 2 hours 43 min | Completed* | 3 hours 1 min | Completed |
3 | CI/CO + DI MARC BIB Create (20 users CI/CO, 25k records DI on 3 tenants) | PTF - Create 2 | 24 min 18 sec | Completed | 31 min 31 sec | Completed | 24 min | Completed * |
4 | 100K MARC BIB Update (Create new file) | PTF - Updates Success - 1 | 58 min 25 sec 57 min 19 sec | Completed | 1 hour 3 min | Completed | - | - |
4 | 250K MARC BIB Update | PTF - Updates Success - 1 | 2 hours 2 min **
2 hours 12 min | Completed with errors ** Completed | 1 hour 53 min | Completed | - | - |
4 | 500K MARC BIB Update | PTF - Updates Success - 1 | 4 hours 43 min 4 hours 38 minutes | Completed Completed | 5 hour 59 min | Completed | - | - |
* - One record on one tenant could be discarded with error: io.netty.channel.StacklessClosedChannelException. https://folio-org.atlassian.net/browse/MODDATAIMP-748 Reproduces in both cases with and without splitting features in at least 30% of test runs with 500k record files and multitenant testing.
** - up to 10 items were discarded with the error: io.vertx.core.impl.NoStackTraceThrowable: Cannot get actual Item by id: org.folio.inventory.exceptions.InternalServerErrorException: Access for user 'data-import-system-user' (f3486d35-f7f7-4a69-bcd0-d8e5a35cb292) requires permission: inventory-storage.items.item.get. Less than 1% of records could be discarded due to missing permission for 'data-import-system-user'. Permission was not added automatically during the service deployment. I added permission manually to the database and the error does not occur anymore.https://folio-org.atlassian.net/browse/MODDATAIMP-930
Test 1,2. 100k, 250K, 500k and Multitenant MARC BIB Create
Memory Utilization
This has memory utilization increasing due to previous modules restarting (everyday cluster shot down process) no memory leak is suspected for DI modules.
MARC BIB CREATE
Test#1 100k, 250k, 500k records DI
Test#2 Multitenant DI (9 concurrent jobs)
Service CPU Utilization
MARC BIB CREATE
Average CPU usage for mod-inventory -was 144%, mod-di-converter-storage was about 107%, and for all other modules did not exceed 100 %. We can observe spikes in CPU usage of mod-data-import at the beginning of the Data Import jobs up to 260%.
Test#1 500k records DI
Test#2 Multitenant
Instance CPU Utilization
Test#1 500k records DI
Test#2 Multitenant DI (9 concurrent jobs)
RDS CPU Utilization
MARC BIB CREATE
Approximately DB CPU usage is up to 95%
Test#1 500k records DI
Test#2 Multitenant DI (9 concurrent jobs)
Maximal DB CPU usage is about 95%
RDS Database Connections
MARC BIB CREATE
For DI job Create- Maximum 535 connections count.
Test#1 500k records DI
Test#2 Multitenant
Test 3 With CI/CO 20 users and DI 25k records on each of the 3 tenants Splitting Feature Enabled &
Splitting Feature Disabled
| Response time without DI Before Splitting Feature Deployed | Response time with DI Before Splitting Feature Deployed | Response time without DI Splitting Feature disabled | Response time with DI Splitting Feature disabled | Response time without DI Splitting Feature enabled | Response time with DI (Average) Splitting Feature enabled |
|---|---|---|---|---|---|---|
Check-In | 0.517s | 1.138s | 0.542s | 1.1s | 0.505s | 1.067s |
Check-Out | 0.796s | 1.552s | 0.841s | 1.6s | 0.804s | 1.48s |
| DI Duration without CI/CO Before Splitting Feature Deployed | DI Duration with CI/CO Before Splitting Feature Deployed | DI Duration without CI/CO Splitting Feature disabled | DI Duration with CI/CO Splitting Feature disabled | DI Duration without CI/CO | DI Duration with CI/CO |
|---|---|---|---|---|---|---|
Tenant _1 | 14 min (18 min for run 2) | 20 min | 27min 47sec | 31min 30sec | 16min 18sec | 16 min 53 sec |
Tenant _2 | 16 min (18 min for run 2) | 19 min | 23min 16sec | 26min 22sec | 20min 13sec | 20min 39 sec |
Tenant _3 | 16 min (15 min for run 2) | 16 min | 18min 40sec | 20min 44sec | 17min 42sec | 17min 54 sec |
* - Same approach testing DI: 3 DI jobs total on 3 tenants without CI/CO. Start the second job after the first one reaches 30%, and start another job on a third tenant after the first job reaches 60% completion. DI file size: 25k
Response time graph
With CI/CO 20 users and DI 25k records on each of the 3 tenants Splitting Feature Disabled
ocp3-mod-data-import:12
Data Import Robustness Enhancement
25K records | RECORDS_PER_SPLIT_FILE | ||||||||||
Number of concurrent tenants | Job profile | 500 | Status | 1K | Status | 5K | Status | 10K | Status | Test with Split disabled | Status |
|---|---|---|---|---|---|---|---|---|---|---|---|
1 Tenant test#1 | PTF - Create 2 | 12 minutes 55 seconds | Completed | 11 minutes 48 seconds | Completed | 09 minutes 21 seconds | Completed | 9 minutes 2 sec | Completed | 10 minutes 35 sec | Completed |
1 Tenant test#2 | 10 minutes 31 seconds | Completed | 09 minutes 32 seconds | Completed | 9 minutes 6 sec | Completed | 9 minutes 14 sec | Completed | 11 minutes 27 sec | Completed | |
2 Tenants test#1 | PTF - Create 2 | 19 minutes 29 seconds | Completed | 15 minutes 47 seconds | Completed | 16 minutes 15 seconds | Completed | 16 minutes 3 seconds | Completed | 19 minutes 18 sec | Completed |
2 Tenants test#2 | 18 minutes 19 seconds | Completed | 15 minutes 47 seconds | Completed | 16 minutes 11 sec | Completed | 16 min 41 sec | Completed | 20 minutes 33 sec | Completed | |
3 Tenants test#1 | PTF - Create 2 | 24 minutes 15 seconds | Completed | 25 minutes 47 seconds | Completed | 23 minutes | Completed | 23 minutes 27 seconds | Completed | 30 minutes 2 sec | Completed |
3 Tenants test#2 | 24 minutes 38 seconds | Completed | 23 minutes 28 seconds | Completed | 23 minutes 2 sec | Completed | 23 minutes 26 seconds | Completed | 29 minutes 54 sec | Completed * | |
* T1 - "00:33:35.1" Error T2 - "01:23:36.144" T3 - "01:16:26.391" on the first tenant proccesing stoped wit error "io.vertx.core.impl.NoStackTraceThrowable: Connection is not active now, current status: CLOSED "
it caused the spike of CPU utilization on Kafka (tenant cluster) up to 94%
Instance CPU Utilization
Test 1. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 500, 2 runs for each test. The maximal CPU Utilization value is 38%.
Test 2. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 10K, 2 runs for each test. The maximal CPU Utilization value is 37%.
Memory Utilization
Test 1. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 500, 2 runs for each test.
Most of the modules were stable during the test, and no memory leak is suspected for DI modules, only 2 modules increased memory consumption usage after the beginning of the tests
Memory utilization rich maximal value for mod-source-record-storage-b 88% and for mod-source-record-manager-b 85%.
Test 2. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 10K, 2 runs for each test.
Service CPU Utilization
Test 1. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 500, 2 runs for each test.
Test 2. Test with 1, 2, and 3 tenants' concurrent jobs with configuration RECORDS_PER_SPLIT_FILE = 10K, 2 runs for each test.
CPU utilization of mod-di-converter-storage-b